Blurts

The conflict between OpenAI’s Board of Directors and Sam Altman was a complex and controversial event. Altman was initially fired by the Board, but then reinstated after several Board members resigned. A New Yorker article provides a detailed account of the situation, including the role of the OpenAI Charter and the different perspectives of Altman and the Board. Educators may be interested in this controversy because OpenAI is responsible for ChatGPT and this service has received a good deal of classroom attention. For some, the OpenAI controversy led to concerns regarding whether ChatGPT was an appropriate tool.

The Charter of OpenAI states that the organization’s mission is to ensure that artificial general intelligence benefits all of humanity and is to be developed in a careful and safe manner. The Board felt that Altman was not adhering to this mission and that he was prioritizing the development of AI over safety concerns. Altman, on the other hand, argued that the Board was being too cautious and that the only way to make progress in AI was to move quickly.

Fast progress required funding which Altman and developers thought could be provided by sale of focused ChatGPT applications and copilot development in collaboration with Microsoft. OpenAI also developed a relationship with Microsoft. Microsoft initially invested one billion and now a total of 13 billion.

Microsoft was interested in the collaboration to support the development of copilot and differences in how AI progress should be achieved would not exactly meet the expectations of careful and safe. Microsoft’s way of thinking about copilot and AI capabilities provided the public reasoned that what was made available would not have to be perfect and users would understand this was the case. Users would understand AI recommendations should be treated as suggestions that require evaluation and these same users would report when recommendations were flawed. Deployment “in the wild” under such circumstances was the best way to discover problems the developers could then fix. 

The lack of trust that was initially given as an explanation for Altman’s termination was reported in the article to be related to a conflict between Altman and Board member Helen Toner. Toner had written a paper critical of OpenAI. In comparing notes, Board members discovered that Altman had been making claims that members had suggested getting rid of Toner and this was evidently not the case (more on the Toner conflict). 

The New Yorker article provides a valuable perspective on the OpenAI conflict. It sheds light on the different factors that contributed to the conflict, and it offers insights into the different ways in which AI can be developed and used.

The article also raises important questions about the future of AI. How can we ensure that AI is developed in a way that benefits all of humanity? How can we balance the need for safety with the need for progress? These are complex questions, but they are essential to the future of AI.

The New Yorker article is a valuable resource for anyone who is interested in the future of AI. It provides a balanced and comprehensive account of the OpenAI conflict, and it raises important questions about the development and use of AI.

Several of the social media outlets I follow are asking for money claiming that ad revenue is drying up. Wikipedia does not take advantage of ad revenue, but is also asking for donations to continue their present approach. I link to Wikipedia often enough within my posts I decided it was appropriate I donate. I would argue many others should consider their own online activity and come to the same conclusion.

I have content on both Medium and Substack. As a writer, I do not regard either platform with the seriousness required to develop a significant following or assume I should expect compensation for my contributions.  I have been a blogger since 2002 and do not see abandoning this body of work and focusing on either Substack and Medium. However, experience with both platforms has given me insights into differences in the experiences of readers and these are my focus in this post.

The cost to readers depends on what a reader wants. If you are looking for unlimited free content, you can find lots of free posts on Substack. Medium limits free access to three articles per month. A subscription to Medium costs $50 per year, but then provides unlimited access to the content generated by all authors. Authors wanting to earn money from their work on Substack typically offer some content for free, but also put content behind a subscription paywall. The balance between free and paid is up to the author. Subscriptions to the work of individual authors vary in annual cost, but $50 is probably a reasonable estimate for purposes of platform comparisons.

I think the choice comes down to how much value you place in the work of an individual author. I think about a $50 subscription in comparison to how many Kindle books I could buy for this investment. I purchase probably one Kindle book a month and I also have an audiobook annual subscription that also requires the purchase of one book a month. I find the lengthy consideration of an issue provided by a single author integrated as a book preferable to following say the work of a blogger who likely mixes multiple topics over an extended period of time. I can purchase three Kindle books for each Substack author I might follow. For the same expenditure, I can read as many of the Medium posts from any writer I want to follow. 

I have no idea what the monetary advantages to an author would be on Medium versus Substack. I am certain individuals who seek revenue from their posts have examined this issue carefully. My guess would be that “popular” writers would prefer Substack and those wanting to make some money but who lack a large following would prefer Medium. 

A more interesting question is what do readers prefer. There is plenty to read on Substack without a subscription to a Substack author, but for the cost of one Substack subscription I can read the work of all Medium contributors. Part of the issue here is what is the quality or value of what is available on Medium and how does a reader reach a decision about the value of complete access when a prospective reader is allowed to review a very limited number of offerings each month. 

I have a couple of subscriptions on Substack and I have a Medium subscription. I like the idea of micropayments which is what Medium appears to support. Writers receive more revenue when more readers access and read their work. In comparison to an ad-based model, I like the micropayment approach. 

I always run into a problem of the consideration of multiple characteristics when I am making a decision regarding online services. Often, each option has some specific things I like and some I do not. This is a good example of such a conundrum. Maybe I should turn the issue around a make statement to authors. I like some of your content, but not well enough to add you to those I subscribe to on Substack. For me, you would get some of my “reading money” if you put your content on Medium.

I should add one caveat. I think I understand the pricing options offered by these two platforms. As a writer, I don’t have enough experience to know for certain. I welcome corrections if I have something wrong.

I recently read what I found to be a depressing account of the future of AI by Deepmind founder Mustafa Suleyman (The Coming Wave). The author examines developments in synthetic biology and artificial intelligence concluding that despite the obvious advantages of both fields (and the interaction of these two technologies) both areas are advancing in capabilities and applications so rapidly that serious dangers are very difficult to address. 

When I say depressing, I mean that ways to address the negative potential of these fields seem unavoidable. His explanation of the dangers made a solid case. Both fields have escaped the typically slow pace of academic research labs and are being developed by commercial entities. The money required for the infrastructure and personnel requirements are beyond the reach of even the most elite universities. These organizations are in a breakneck race to control the market with huge revenue opportunities in the balance. Whether or not our government will step in to establish regulations and monitoring to require careful consideration of dangers, the governments in other countries will not have the same motives and see these fields as key to international power. 

While bigger is one of the characteristics creating this challenge, smaller also contributes. While large corporations or even nations create the tools of AI and synthetic biology, the use of these tools once available is in control of individuals. Many will have the capabilities to implement uses of AI and synthetic biology.

The recent OpenAI upheaval seems an example of some of these issues in action. OpenAI originated in part to provide an alternative to the efforts of big corporations. Even as a well-funded nonprofit, OpenAI sought additional funding opportunities in part leading to disagreements between the board and key personnel. It seemed AI innovation would continue no matter any position taken by OpenAI and just to remain a credible option to the efforts of huge established companies, OpenAI must sell products and make its products more adaptable to the interests of individual users.

It has been a crazy week for OpenAI and ChatGPT. Sam Altman announces customizable GPTs and then he gets let go by his board. I have no idea if these events are related and information is scarce. It appears that Altman may be negotiating with the board so it is difficult to know what to expect.

I am very interested in the new GPT opportunities. I have found applying an AI client to investigate content I specify to be one of the most immediate opportunities and this appears to be the benefit of some of the customizations. You do need to pay for the $20 a month plan to apply any of these options. When I tried I found that the system is overloaded and interested users are now on a waiting list.